EQUIVALENCE CLASS IN THE SET OF FUZZY NUMBERS

AND ITS APPLICATION IN DECISION-MAKING PROBLEMS

GEETANJALI PANDA, MOTILAL PANIGRAHI, AND SUDARSAN NANDA
Received 4 November 2005; Revised 20 June 2006; Accepted 22 June 2006
An equivalence relation is dened in the set of fuzzy numbers. In a particular equivalence
class, arithmetic operations of fuzzy numbers are introduced. A fuzzy matrix with respect
to a particular class and its associated crisp matrices are also introduced. The concept of
equivalence class is applied in fuzzy decision-making problems and justied through a
numerical example.
Copyright 2006 Hindawi Publishing Corporation. All rights reserved.
1. Introduction
Since the discovery of fuzzy sets, the arithmetic operations of fuzzy numbers (Zadeh
[7, 8]) which may be viewed as a generalization of interval arithmetic (Moore [5]) have
emerged as an important area of research within the theory of fuzzy sets (Mizumoto and
Tanaka [4], Dubois and Prade [2]). The arithmetic operations of fuzzy numbers have
been performed either by extension principle [7, 8] or by using alpha cuts as discussed by
Dubois and Prade [2]. If two triangular (linear) fuzzy numbers are added or subtracted
by applying extension principle, then the result is again a triangular fuzzy number. How-
ever, when we take other operations like multiplication or division, then the result is
not a linear triangular fuzzy number. Thus these operations are not closed in the sense
that operations of two same types (triangular/trapezoidal, etc.) of fuzzy numbers may
not necessarily result in fuzzy number of that type (triangular/trapezoidal, etc.). Hence it
is cumbersome to nd the membership function of the arithmetical operations of large
number of fuzzy numbers (may be of same type) by these principles. In the present paper,
an attempt is made to overcome such type of diculties. In Section 3, a relation between
two fuzzy numbers is dened and it has been proved that this relation divides the whole
set of fuzzy numbers into equivalence classes. In Section 4, arithmetic operations in a par-
ticular class are dened. These operations are dierent from the arithmetic operations of
fuzzy numbers, developed by extension principle or alpha cut. It has been proved and also
veried through examples that these arithmetic operations are closed in their respective
Hindawi Publishing Corporation
International Journal of Mathematics and Mathematical Sciences
Volume 2006, Article ID 74165, Pages 119
DOI 10.1155/IJMMS/2006/74165
2 Equivalence class in fuzzy numbers
classes (e.g., product of two triangular fuzzy numbers in a particular class results in a
triangular fuzzy number of the same class).
[]-fuzzy matrix and arithmetic operations of []-fuzzy matrices are introduced in
Section 5. Some properties such as deniteness, symmetry of []-fuzzy matrices (these
are useful to formulate quadratic programming) are discussed.
To make these concepts of equivalence class, arithmetic operations in a particular
class and []-fuzzy matrix useful in practical sense, an attempt is made to apply it in
fuzzy decision-making problems. Fuzzy decision making was discovered by Bellman and
Zadeh [1] and a lot of work has been done by many researchers like Zimmermann [9],
Werners [6], and so forth to solve fuzzy decision-making problems using dierent types
of defuzzication methods. In this paper, a dierent type of solution method for fuzzy
decision-making problem has been developed without defuzzifying it. In any defuzzica-
tion method, solution of a fuzzy decision-making problem does not involve the aspira-
tion level of the fuzzy numbers present in the problem. But in our solution method, each
aspiration level corresponds to a solution. We have considered a fuzzy decision-making
problem whose coecients are fuzzy numbers belonging to a particular class. A general
fuzzy decision-making problem in matrix form is formulated and its solution method is
given in Section 6. The methodology is veried for fuzzy quadratic programming prob-
lem.
Throughout this paper, we have considered the set of all fuzzy numbers with compact
support and also LR type.
2. Some preliminary results on fuzzy numbers
Denition 2.1 (fuzzy number). A fuzzy number is a fuzzy subset of the real line R which
is convex and normal.
Let F be the set of all fuzzy numbers which are upper semicontinuous and have com-
pact support. The support of a fuzzy number ` a with membership function
` a
: R[0, 1]
is denoted by supp(` a), where supp(` a) =[l
a
, r
a
] and

` a
(t) =

` aL
(t), l
a
t a

,
1, a

t a

` aR
(t), a

t r
a
,
0, otherwise.
(2.1)

` aL
: [l
a
, a

] [0, 1] is a continuous and strictly increasing function and

` aR
: [a

, r
a
]
[0, 1] is a continuous and strictly decreasing function.
` a
(t) = 1 for t [a

, a

] and is
called the core of ` a, denoted by core(` a). Symbolically, the fuzzy number ` a with compact
support is represented by ` a =l
a
; a

; a

; r
a
).
Denition 2.2 (triangular fuzzy number). A fuzzy number ` a =l
a
; a

; a

; r
a
) is said to be
a triangular fuzzy number if a

= a

. Moreover, if the membership function

` a
is such
that
` aL
and
` aR
are linear, then ` a is a linear triangular fuzzy number.
Geetanjali Panda et al. 3
Denition 2.3 (LR fuzzy number). A fuzzy number ` a is of LR type if there exist reference
functions L and R, scalars > 0, > 0 with

` a
(t) =

L
_
a

_
, t a

,
1, a

t a

,
R
_
t a

_
, t a

.
(2.2)
[a

, a

] is the core of ` a, and are the left and right spreads, respectively. L, R : R
+

[0, 1] are decreasing shape functions. L(0) =R(0) =1, L(1) =R(1) =0 (or L(t) > 0, R(t) >
0, for all t, L() = R(+) = 0). An LR-type fuzzy number ` a is represented by ` a =
a

; a

; , )
LR
. Let F

denote the set of all LR fuzzy numbers.

Denition 2.4 (ranking of fuzzy numbers [3]). For 0 1, the -integral value of a
fuzzy number ` a, denoted by I

(` a), is
I

(` a) =R(` a) +(1 )L(` a), (2.3)

where
R(` a) =
_
1
0

1
` aR
()d,
L(` a) =
_
1
0

1
` aL
()d.
(2.4)
For two fuzzy numbers ` a and
`
b, ` a i
`
b (less than or equal to in fuzzy sense) if and only if
I

(` a) I

(
`
b) for 0 1.
The next section is devoted to dene equivalence class in the set of fuzzy numbers.
3. Equivalence class in fuzzy numbers
Before introducing an equivalence relation, we dene the domain of a fuzzy number in F
and F

as follows.
(I) Suppose ` a =l
a
; a

; a

; r
a
) and
`
b =l
b
; b

; b

; r
b
) are two fuzzy numbers in F with
membership functions
` a
and
`
b
, respectively. The domains of ` a and
`
b are D
` a
and D
`
b
,
respectively, which are real lines and can be represented by
D
` a
=
_
, l
a
_

_
l
a
, r
a
_

_
r
a
,
_
,
D
`
b
=
_
, l
b
_

_
l
b
, r
b
_

_
r
b
,
_
.
(3.1)
4 Equivalence class in fuzzy numbers
Let f : D
` a
D
`
b
be dened as
f (t) =

t +l
b
l
a
, t l
a
,
b

_
t l
a
_
+l
b
_
a

t
_
a

l
a
, l
a
t a

,
b

_
t a

_
+b

_
a

t
_
a

, a

t a

,
r
b
_
t a

_
+b

_
r
a
t
_
r
a
a

, a

t r
a
,
t +r
b
r
a
, t r
a
.
(3.2)
(II) Suppose ` a and
`
b are in F

, ` a =a

; a

; , )
LR
, and
`
b =b

; b

; , )
LR
. Then
D
` a
=
_
, a

_
a

, a

+
_

_
a

+,
_
,
D
`
b
=
_
, b

_
b

, b

+
_

_
b

+,
_
.
(3.3)
In this case, f : D
` a
D
`
b
can be dened as in (3.2), just by replacing l
a
by a

, l
b
by
b

, r
a
by a

+, and r
b
by b

+.
It is not hard to see that f is bijective in both cases.
Denition 3.1. For ` a,
`
bF (or F

), dene a relation

= between ` a and
`
b (
`
b is said to be
related to ` a by the relation

=) as
` a

=
`
b i
` a
(t) =
`
b
_
f (t)
_
. (3.4)
Example 3.2. Let ` a1; 2; 4),
`
b2; 5; 7) F be two linear triangular fuzzy numbers. Then

` a
(t) =

t 1, 1 t 2,
2
t
2
, 2 t 4,
0, otherwise,

`
b
(t) =

t 2
3
, 2 t 5,
7 t
2
, 5 t 7,
0, otherwise,
f (t) =

t +1, t 1,
3t 1, 1 t 2,
t +3, t 2.
(3.5)
For all t,
` a
(t) =
`
b
( f (t)). Hence ` a

=
`
b.
Geetanjali Panda et al. 5
Example 3.3. ` a = 2; 1, 2)
L

R
,
`
b = 5; 1, 2)
L

R
are two fuzzy numbers in F

. L

(t) =
(1 +t
2
)
1
, R

(t) =e
t
. Then

` a
(t) =

1
1 +(2 t)
2
, t 2,
e
(t2)/2
, t 2,

`
b
(t) =

1
1 +(5 t)
2
, t 5,
e
(t5)/2
, t 5,
(3.6)
f (t) =t +3 and
` a
(t) =
`
b
(t +3). Hence ` a

=
`
b.
Theorem 3.4.

= is an equivalence relation in F.
Proof. Let ` a,
`
b, ` cF, ` a =l
a
; a

; a

; r
a
),
`
b =l
b
; b

; b

; r
b
), and ` c =l
c
; c

; c

; r
c
).
(i) Obviously,

= is reexive.
(ii) Let ` a

=
`
b, that is,
` a
(t) =
`
b
( f (t)), where f is dened as in (3.2). Since f is bijec-
tive, f
1
: D
`
b
D
` a
exists and is dened as
f
1
(t) =

t +l
a
l
b
, t l
b
,
a

_
t l
b
_
+l
a
_
b

t
_
b

l
b
, l
b
t b

,
a

_
t b

_
+a

_
b

t
_
b

, b

t b

,
r
a
_
t b

_
+a

_
r
b
t
_
r
b
b

, b

t r
b
,
t +r
a
r
b
, t r
b
.
(3.7)
It is easy to see that f
1
( f (t)) = f f
1
(t) =t for each t. Also

` a
( f
1
(t)) =

` aL
_
f
1
(t)
_
, l
a
f
1
(t) a

,
1, a

` aR
_
f
1
(t)
_
, a

f
1
(t) r
a
,
0, otherwise
6 Equivalence class in fuzzy numbers
=

`
bL
_
b

_
f
1
(t) l
a
_
+l
b
_
a

f
1
(t)
_
a

l
a
_
, l
a
f
1
(t) a

,
1, a

f
1
(t) a

`
bR
_
r
b
_
f
1
(t) a

_
+b

_
r
a
f
1
(t)
_
r
a
a

_
, a

f
1
(t) r
a
,
0, otherwise
=

`
bL
(t), l
b
t b

,
1, b

t b

`
bR
(t), b

t r
b
,
0, otherwise
=
`
b
(t).
(3.8)
Hence
`
b

= ` a, that is,

= is symmetric.
(iii) Let ` a

=
`
b and
`
b

= ` c. f : D
` a
D
`
b
is as dened in (3.2) and g : D
`
b
D
` c
is
g(t) =

t +l
c
l
b
, t l
b
,
c

_
t l
b
_
+l
c
_
b

t
_
b

l
b
, l
b
t b

,
c

_
t b

_
+c

_
b

t
_
b

, b

t b

,
r
c
_
t b

_
+c

_
r
b
t
_
r
b
b

, b

t r
b
,
t +r
c
r
b
, t r
b
,
(3.9)
such that
` a
(t) =
`
b
( f (t)),
`
b
(t) =
` c
(g(t)). So
` a
(t) =
` c
(g( f (t))) =
` c
((g
o
f )(t)), where
g
o
f : D
` a
D
` c
is
g
o
f (t) =

f (t) +l
c
l
b
, f (t) l
b
,
c

_
f (t) l
b
_
+l
c
_
b

f (t)
_
b

l
b
, l
b
f (t) b

,
c

_
f (t) b

_
+c

_
b

f (t)
_
b

, b

f (t) b

,
r
c
_
f (t) b

_
+c

_
r
b
f (t)
_
r
b
b

, b

f (t) r
b
,
f (t) +r
c
r
b
, f (t) r
b
,
Geetanjali Panda et al. 7
=

t +l
c
l
a
, t l
a
,
c

_
t l
a
_
+l
c
_
a

t
_
a

l
a
, l
a
t a

,
c

_
t a

_
+c

_
a

t
_
a

, a

t a

,
r
c
_
t a

_
+c

_
r
a
t
_
r
a
a

, a

t r
a
,
t +r
c
r
a
, t r
a
.
(3.10)
Hence ` a

= ` c, that is,

= is transitive.
So

= is an equivalence relation in F.
Theorem 3.5.

= is an equivalence relation in F

.
Proof. Proof of this result is similar to the proof of Theorem 3.4.
The set of all the fuzzy numbers in F (or F

), which are equivalent to each other by

the relation

=, is known as an equivalence class and we denote it by []. (In fact, the
equivalence class [] contains all the fuzzy numbers with similar membership function
.) Thus [] may be a triangular class, trapezoidal class, Gaussian class or a particular
type of LR class, and so forth.
4. Arithmetic operations of fuzzy numbers in an equivalence class
Suppose ` a,
`
b [], ` a =l
a
; a

; a

; r
a
), and
`
b =l
b
; b

; b

; r
b
) [],
supp(` a) =
_
l
a
, r
a
_
, core(` a) =
_
a

, a

_
,
domain of ` a =D
` a
=
_
, l
a
_

_
l
a
, r
a
_

_
r
a
,
_
,
supp(
`
b) =
_
l
b
, r
b
_
, core(
`
b) =
_
b

, b

_
,
domain of
`
b =D
`
b
=
_
, l
b
_

_
l
b
, r
b
_

_
r
b
,
_
.
(4.1)
We now dene the arithmetic operation as follows:
` a
`
b = ` c, (4.2)
where the fuzzy number ` c is dened as below:
supp(` c) =supp(` a)

supp(
`
b) =
_
l
a
, r
a
_

_
l
b
, r
b
_
=
_
l
c
, r
c
_
,
core(` c) =core(` a)

core(
`
b) =
_
a

, a

_
b

, b

_
=
_
c

, c

_
.
(4.3)
Here

denotes the arithmetic operation of intervals corresponding to fuzzy arithmetic

operation (i.e., if represents fuzzy multiplication, then

represents multiplication
8 Equivalence class in fuzzy numbers
of intervals). The domain of ` c denoted by D
` c
or D
` a
`
b
is
D
` c
=
_
, l
c
_

_
l
c
, r
c
_

_
r
c
,
_
. (4.4)
Theorem 4.1. If ` a,
`
b [], then ` a
`
b [].
Proof. Let ` a,
`
b [] and ` a
`
b = ` c is dened as above.
We dene g : D
` a
D
` a
`
b
by
g(t) =

t +l
c
l
a
, t l
a
,
c

_
t l
a
_
+l
c
_
a

t
_
a

l
a
, l
a
t a

,
c

_
t a

_
+c

_
a

t
_
a

, a

t a

,
r
c
_
t a

_
+c

_
r
a
t
_
r
a
a

, a

t r
a
,
t +r
c
r
a
, t r
a
.
(4.5)
The membership value of the fuzzy number ` a
`
b is dened as

` a
`
b
_
g(t)
_
=
` a
(t). (4.6)
Since g is a bijective mapping, so

` c
(t) =
` a
`
b
(t) =
` a
_
g
1
(t)
_
. (4.7)
Hence ` a

= ` a
`
b, that is, ` a
`
b []. This proves that is closed in [].
Let us consider an example of fuzzy multiplication. We denote the multiplication of
two fuzzy numbers by .
Example 4.2 (multiplication). Consider the following two fuzzy numbers: ` a =1; 3; 6; 7),
`
b =2; 3; 5; 8) []. [] is a linear trapezoidal class. Then the membership functions of ` a
and
`
b are

` a
(t) =

t 1
2
, 1 t 3,
1, 3 t 6,
7 t, 6 t 7,
0, otherwise,

`
b
(t) =

t 2, 2 t 3,
1, 3 t 5,
8 t
3
, 5 t 8,
0, otherwise,
(4.8)
Geetanjali Panda et al. 9
supp(` a
`
b) =[2, 56], core(` a
`
b) =[9, 30]. Now
g(t) =

t +1, t 1,
7t 3
2
, 1 t 3,
7t 12, 3 t 6,
26t 126, 6 t 7,
t +49, t 7.
(4.9)
We will evaluate
` a
`
b
as below.
(i) t 2 g
1
(t) 1. So
` a
`
b
(t) =
` a
(g
1
(t)) =0.
(ii) 2 t 9 1 g
1
(t) 3. So
` a
`
b
(t) =
` a
(g
1
(t)) = (g
1
(t) 1)/2 = ((2t +3)/
7 1)/2 =(t 2)/7.
(iii) 9 t 30 3 g
1
(t) 6. So
` a
`
b
(t) =
` a
(g
1
(t)) =1.
(iv) 30 t 56 6 g
1
(t) 7. So
` a
`
b
(t) =
` a
(g
1
(t)) =7 g
1
(t)=7 (t +126)/
26 =(56 t)/26.
(v) t 56 g
1
(t) 7. So
` a
`
b
(t) =
` a
(g
1
(t)) =0.
On the other hand, we can dene h : D
`
b
D
` a
`
b
by
h(t) =

t, t 2,
7t 12, 2 t 3,
21t 45
2
, 3 t 5,
26t 40
3
, 5 t 8,
t +48, t 8.
(4.10)
(i) t 2 h
1
(t) 2. So
` a
`
b
(t) =
`
b
(h
1
(t)) =0.
(ii) 2 t 9 2 h
1
(t) 3. So
` a
`
b
(t) =
`
b
(h
1
(t)) = h
1
(t) 2 = (t + 12)/7
2 =(t 2)/(7).
(iii) 9 t 30 3 h
1
(t) 5. So
` a
`
b
(t) =
`
b
(h
1
(t)) =1.
(iv) 30 t 56 5 h
1
(t) 8. So
` a
`
b
(t) =
`
b
(h
1
(t)) = (8 h
1
(t))/3 = (8
(3t +40)/26)/3 =(56 t)/26.
(v) t 56 h
1
(t) 8. So
` a
`
b
(t) =
`
b
(h
1
(t)) =0.
From both cases, it is clear that

` a
`
b
(t) =

t 2
7
, 2 t 9,
1, 9 t 30,
56 t
26
, 30 t 56,
0, otherwise.
(4.11)
10 Equivalence class in fuzzy numbers
This is the membership function of the fuzzy number ` a
`
b =2; 9; 30; 56). Hence ` a
`
b
[].
To compare this arithmetic operation with the arithmetic operation by using extension
principle, we state the following example.
Example 4.3. Consider a linear triangular class []. Suppose ` a = 1; 2; 4),
`
b = 2; 5; 7)
[] are two linear triangular fuzzy numbers. The membership functions are given by

` a
(t) =

t 1, 1 t 2,
2
t
2
, 2 t 4,
0, otherwise,

`
b
(t) =

t 2
3
, 2 t 5,
7 t
2
, 5 t 7,
0, otherwise.
(4.12)
(I) The multiplication of ` a and
`
b using the arithmetic operation as developed in Section 4.
We have supp(` a
`
b) =[2, 28], core(` a
`
b) =10, and g : D
` a
D
` a
`
b
,
g(t) =

t +1, t 1,
8t 6, 1 t 2,
9t 8, 2 t 4,
t +24, t 4.
(4.13)
Now we calculate
` a
`
b
.
(i) t 2 g
1
(t) 1. So
` a
`
b
(t) =
` a
(g
1
(t)) =0.
(ii) 2 t 10 1 g
1
(t) 2. So
` a
`
b
(t) =
` a
(g
1
(t)) = g
1
(t) 1 = (t + 6)/8
1 =(t 2)/8.
(iii) 10 t 28 2 g
1
(t) 4. So
` a
`
b
(t) =
` a
(g
1
(t)) = 2 g
1
(t)/2 = 2 (t +
8)/18 =(28 t)/18.
(iv) t 28 g
1
(t) 4. So
` a
`
b
(t) =
` a
(g
1
(t)) =0.
On the other hand, we can dene h : D
`
b
D
` a
`
b
by
h(t) =

t, t 2,
8t 10
3
, 2 t 5,
9t 35, 5 t 7,
t +21, t 7,
(4.14)
and again calculate
` a
`
b
.
(i) t 2 h
1
(t) 2. So
` a
`
b
(t) =
`
b
(h
1
(t)) =0.
(ii) 2 t 10 2 h
1
(t) 5. So
` a
`
b
(t) =
` a
(h
1
(t))=(h
1
(t) 2)/3 =((3t +10)/
(8 2))/3 =(t 2)/8.
Geetanjali Panda et al. 11
(iii) 10 t 28 5 h
1
(t) 7. So
` a
`
b
(t) =
` a
(h
1
(t)) =(7 h
1
(t))/2 =(7 (t +
35))/9/2 =(28 t)/18.
(iv) t 28 h
1
(t) 4. So
` a
`
b
(t) =
` a
(h
1
(t)) =0.
Thus from both cases, it is clear that

` a
`
b
(t) =

t 2
8
, 2 t 10,
28 t
18
, 10 t 28,
0, otherwise.
(4.15)
This is the membership function of the linear triangular fuzzy number ` a
`
b =2; 10; 28).
(II) The multiplication of ` a and
`
b using the extension principle.
By the extension principle of the arithmetic operations of fuzzy numbers, the multi-
plication of the two fuzzy numbers ` a and
`
b with linear triangular membership functions
as given in Example 4.3 would have been the following fuzzy number ` a
`
b say, whose
membership function is given by

` a
`
b
(t) =

1
6
_
12t +1 5
_
, 2 t 10,
1
4
_
11

4t +9
_
, 10 t 28,
0, otherwise.
(4.16)
Here
` a
`
b
is a nonlinear membership function even though the two given fuzzy numbers
were linear triangular. That is, the multiplication of two linear triangular fuzzy numbers
is not a linear triangular fuzzy number by the usual fuzzy multiplication. But in (I), it
has been veried that multiplication of two linear triangular fuzzy numbers is a linear
triangular fuzzy number.
Scalar multiplication. The scalar multiplication between a crisp number and a fuzzy
number ` a =l
a
; a

; a

; r
a
) with
` a
as membership function is done as follows.
(i) If 0, then ` a = l
a
; a

; a

; r
a
), here a map f : D
` a
D
` a
can be dened
as in Theorem 4.1 and the membership function is given by
` a
=
` a
( f
1
(t)).
(ii) If < 0, then ` a = r
a
; a

; a

; l
a
) here again a map f : D
` a
D
` a
can be
dened as in Theorem 4.1 and the membership function is given by
` a
=

` a
( f
1
(t)).
Example 4.4. Consider ` a =1; 3; 6; 7),
`
b =2; 3; 5; 8) as in Example 4.2 and =2, =3.
Then 2` a =2; 6; 12; 14), 3` a =21; 18; 9; 3).
12 Equivalence class in fuzzy numbers
Take ` a = 1; 2; 4) as in Example 4.3 and = 2, = 3. Then 2` a = 2; 4; 8), 3` a =
12; 6; 3), the corresponding membership functions are given below:

2` a
(t) =

t 2
2
, 2 t 4,
8 t
4
, 4 t 8,
0, otherwise,

3` a
(t) =

t +12, 12 t 6,
3 t
3
, 6 t 3,
0, otherwise.
(4.17)
With the above knowledge of equivalence class in the set of fuzzy numbers and the arith-
metic operations in a class, we introduce []-fuzzy matrix in the next section.
5. []-fuzzy matrix
A []-fuzzy matrix of order mn is an arrangement of mn fuzzy numbers of class [] in
m rows and n columns. It is denoted by
`
A

, where
`
A

=(` a
i j
)
mn
, where ` a
i j
[].
Two fuzzy matrices of same class and same order can be added and two fuzzy matrices
of order mn and n p, respectively, in the same class can be multiplied in the sense
of arithmetic operation of fuzzy numbers in an equivalence class as dened in Section 4.
The arithmetic operations such as addition, multiplication of two []-fuzzy matrices, and
the multiplication of a []-fuzzy matrix with a scalar are dened as follows.
(1) Addition: for the fuzzy matrices
`
A

=(` a
i j
) and
`
B

=(
`
b
i j
) of order mn,
`
A

`
B

=
(` a
i j

`
b
i j
).
(2) Multiplication: for the fuzzy matrices
`
A

= (` a
i j
) of order mn and
`
B

= (
`
b
i j
)
of order n p,
`
A

`
B

=(` c
rs
) is of order m p, where ` c
rs
=(` a
r1

`
b
1s
) (` a
r2

`
b
2s
) (` a
rn

`
b
ns
).
(3) Scalar multiplication: for a scalar ,
`
A

=(` a
i j
),
where is the addition of two fuzzy numbers, is the multiplication of two fuzzy num-
bers. and can be evaluated as the general arithmetic operation dened in Section 4.
Example 5.1. Let
`
A

`
2
`
1
`
2
`
3

,
`
B

`
1
`
1
`
2
`
2

, (5.1)
where
`
2 =1; 2; 4),
`
1 =0; 1; 2),
`
3 =1; 3; 4),
`
2,
`
1,
`
3 [], where [] is the linear triangular
class fuzzy numbers
`
A

`
B

=
_
`
3
`
2
`
4
`
5
_
,
`
A

`
B

=
_
`
4
`
4
`
8
`
8
_
, 2
`
A

=
_
`
4
`
2
`
4
`
6
_
, (5.2)
where in
`
A

`
B

,
`
3 = 1; 3; 6),
`
2 = 0; 2; 4),
`
4 = 2; 4; 8),
`
5 = 2; 5; 8), in
`
A

`
B

,
`
4 =
0; 4; 16),
`
8 =1; 8; 24), and in 2
`
A

,
`
4 =2; 4; 8),
`
6 =2; 6; 8).
Geetanjali Panda et al. 13
Note that
`
4=0; 4; 16),
`
4=2; 4; 8) are two dierent fuzzy numbers since I

(0; 4; 16)) }=
I

(2; 4; 8)). But they are similar and are in same linear triangular class.
Denition 5.2 (symmetric fuzzy matrix). A []-fuzzy matrix
`
A

=(` a
i j
) of order n n is
said to be symmetric if
` a
i j
(t) =
` a
ji
(t) for all i }= j and t R.
In the literature, a lot of defuzzication methods or ranking of fuzzy numbers are
available. Here we have considered the ranking method applied through -integral value
to dene a positive semidenite fuzzy matrix. However, any other ranking method can be
applied.
Denition 5.3. Positive semidenite fuzzy matrix. A []-fuzzy matrix
`
A

of order nn is
said to be positive semidenite if for x R
n
, I

(x
T
`
A

x) 0.
Example 5.4. The following matrix is a symmetric fuzzy matrix of order 3 3.
Let
`
A

`
0
`
2
`
4
`
2
`
3
`
4
`
4
`
4
`
2

, (5.3)
where ` a
i j
[], is the membership function of linear triangular fuzzy numbers in F.
Specically, ` a
11
=
`
0 = 1; 0; 2), ` a
12
=
`
2 = 1; 2; 4), ` a
13
= ` a
31
=
`
4 = 0; 4; 5), ` a
22
=
`
3 =
0; 3; 7), ` a
23
= ` a
32
=
`
4 =1; 4; 8), and ` a
33
=
`
2 =1; 2; 3).
Example 5.5. The following []-matrix is positive semidenite.
Let
`
A

=
_
`
6
`
4
`
4

10
_
, where [] is the linear triangular class,
`
6 =4; 6; 7),
`
4 =3; 4; 5),

10 =
9; 10; 12), and x =[x
1
, x
2
], x
1
, x
2
R.
If either x
1
< 0 or x
2
< 0 but not both, then
x
T
`
A

x =
_
4x
2
1
+10x
1
x
2
+9x
2
2
; 6x
2
1
+8x
1
x
2
+10x
2
2
; 7x
2
1
+6x
1
x
2
+12x
2
2
_
. (5.4)
Here we can see that
4x
2
1
+10x
1
x
2
+9x
2
2
=
_
2x
1
+3x
2
_
2
2x
1
x
2
> 0,
6x
2
1
+8x
1
x
2
+10x
2
2
=
_
6x
1
+

10x
2
_
2
+
_
8 2

60
_
x
1
x
2
> 0,
7x
2
1
+6x
1
x
2
+12x
2
2
=
_
7x
1
+

12x
2
_
2
+
_
6 2

84
_
x
1
x
2
> 0.
(5.5)
Otherwise, when both x
1
and x
2
are either > 0 or < 0, then
x
T
`
A

x =
_
4x
2
1
+6x
1
x
2
+9x
2
2
; 6x
2
1
+8x
1
x
2
+10x
2
2
; 7x
2
1
+10x
1
x
2
+12x
2
2
_
. (5.6)
In this case, it is clear that all the expressions are positive.
14 Equivalence class in fuzzy numbers
Hence I

(x
T
`
A

x) > 0 for all x R.

Therefore,
`
A

is positive semidenite.
Denition 5.6 ([]-crisp matrix). Every []-fuzzy matrix,
`
A

, is associated with a pair of

crisp matrices A

() and A
+

() for each 0 < 1. Dene these as

A

() =
_

i j
_
, where

i j
=
1
` a
i j
L
(),
A
+

() =
_

+
i j
_
, where
+
i j
=
1
` a
i j
R
().
(5.7)
Example 5.7. The []-fuzzy matrix in Example 5.5,
`
A

=
_
`
6
`
4
`
4

10
_
, is associated with two
crisp matrices
A

(0.1) =
_
4.2 3.1
3.1 9.1
_
,
A
+

(0.1) =
_
6.9 4.9
4.9 11.8
_
(5.8)
for =0.1.
Proposition 5.8. If a []-fuzzy matrix is symmetric, then the associated crisp matrices are
also symmetric.
Proof. The proof is easy and we omitted it.
Next section and its subsection are devoted to the application of []-class and []-
fuzzy matrix in decision-making problems.
6. Fuzzy decision-making problem
The general form of a fuzzy decision-making problem is
minimize
`
f (x),
subject to ` g(x) i(r, =)
`
0,
x 0,
(6.1)
where
`
f (x), ` g(x) are fuzzy functions fromR
+
to the set of fuzzy numbers.
A lot of defuzzication methods have been developed so far to solve such type of fuzzy
decision-making problems. In this section, we have presented the following algorithm to
solve the problem (6.1) without defuzzifying and using the concept of fuzzy equivalence
class as developed in previous sections.
Suppose that the coecients of
`
f (x), ` g(x) are fuzzy numbers belonging to []-class
and
`
0 [].
Geetanjali Panda et al. 15
Step 1. For a particular aspiration level [0, 1], problem (6.1) can be decomposed to
the following two decision-making problems using []-crisp matrix concept:
minimize
`
f

(, x),
subject to ` g

(, x) (=, )
`
0

(),
0 1,
x 0,
minimize
`
f
+

(, x),
subject to ` g
+

(, x) (=, )

+
(),
0 1,
x 0.
(6.2)
For a particular aspiration level (0, 1), we use (6.2) to get the solution of (6.1).
Step 2. Formulate the corresponding multiobjective problem:
minimize
_
`
f

(, x),
`
f
+

(, x)
_
,
subject to ` g

(, x) (=, )
`
0

(),
` g
+

(, x) (=, )
`
0
+

(),
0 1,
x 0.
(6.3)
Step 3. The multiobjective problem (6.3) can be solved for each aspiration level by any
multiobjective programming method.
This algorithm can be explained through the following fuzzy quadratic programming
problem.
6.1. Fuzzy quadratic programming. The general form of a crisp quadratic program-
ming problem is
minimize x
T
Ax +c
T
x,
subject to Bx (, =)b,
x 0,
(6.4)
where x R
n
, A is a positive semidenite symmetric crisp matrix of order nn, B, c and
b are crisp matrices of order mn, n1, and m1, respectively.
We dene a fuzzy quadratic programming problem as follows:
minimize x
T
`
A

x ` c
T

x,
subject to
`
B

x i(r, =)
`
b

,
x 0,
(6.5)
16 Equivalence class in fuzzy numbers
where x R
n
,
`
A

is a positive semidenite symmetric []-fuzzy matrix of order n n,

`
B

, ` c

, and
`
b

are []fuzzy matrices of order mn, n 1, and m1, respectively. Let

`
f (x) =x
T
`
A

x ` c
T
x and g

(x) =
`
B

x
`
d

, where
`
d

=(1)
`
b

.
Let
`
A

=
_
`
6
`
4
`
4

10
_
, ` c

=
_
`
1

`
2
_
,
`
B

=
_
`
2
`
1
`
1
`
1
_
,
`
b

=
_
`
1
`
4
_
, (6.6)
where [] is the linear triangular class,
`
1 =0; 1; 3),
`
2 =1; 2; 4),
`
4 =3; 4; 5),
`
6 =4; 6; 7),

10 = 9; 10; 12). So
`
2 = 4; 2; 1), x = (x
1
, x
2
)
T
. In Example 5.5, it is veried that
I

(x
T
`
A

x) 0 for all x R
2
and some 0 1. Hence
`
A

is a positive semidenite
[]-fuzzy matrix. Also
`
A

is symmetric. We consider the fuzzy quadratic programming

problem (6.5) as
min
`
6x
2
1
(
`
4
`
4)x
1
x
2

10x
2
2

`
1x
1
(1)
`
2x
2
,
subject to
`
2x
1

`
1x
2
r
`
1,
`
1x
1

`
1x
2
i
`
4,
x
1
, x
2
0.
(6.7)
Solution using the algorithm. For some aspiration level , each of
`
A

, ` c
T

,
`
b

,
`
B

corre-
sponds to a pair of crisp matrices (A

(), A
+

()), (c

(), c
+

()), (b

(), b
+

()),(B

(),
B
+

()).
Hence the fuzzy quadratic programming problem (6.5) corresponds to a pair of crisp
quadratic programming problems
minx
T
A

()x +c

T
()x,
subject to B

()x (=, )b

(),
x 0,
minimize x
T
A
+

()x +c
+

T
()x,
subject to B
+

()x (=, )b
+

(),
x 0,
(6.8)
where
A

() =

2 +4 +3
+3 9 +

, A
+

() =

7 5
5 12 2

, c

() =

2 4

,
c
+

() =

3 2
1

, B

() =

+1

, B
+

() =

4 2 3 2
3 2 3 2

,
b

() =

+3

, b
+

() =

3 2
5

.
(6.9)
Geetanjali Panda et al. 17
Step 1. Substituting the values of the pair of crisp matrices, (6.7) can be converted to two
crisp quadratic programming problems in the light of (6.8) as
minimize (2 +4)x
2
1
+2( +3)x
1
x
2
+(9 +)x
2
2
+x
1
+(2 4)x
2
,
subject to ( +1)x
1
+x
2
,

_
x
1
+x
2
_
+3,
x
1
0, x
2
0, 0 1,
minimize (7 )x
2
1
+2(5 )x
1
x
2
+(12 2)x
2
2
+(3 2)x
1
+(1 )x
2
,
subject to (4 2)x
1
+(3 2)x
2
3 2,
(3 2)
_
x
1
+x
2
_
5 ,
x
1
0, x
2
0, 0 1.
(6.10)
Step 2. The corresponding multiobjective programming is
min
_
(2 +4)x
2
1
+2( +3)x
1
x
2
+(9 +)x
2
2
+x
1
+(2 4)x
2
,
(7 )x
2
1
+2(5 )x
1
x
2
+(12 2)x
2
2
+(3 2)x
1
+(1 )x
2
_
,
subject to ( +1)x
1
+x
2
,
_
x
1
+x
2
_
+3,
(4 2)x
1
+(3 2)x
2
3 2,
(3 2)
_
x
1
+x
2
_
5 ,
x
1
0, x
2
0, 0 1.
(6.11)
Step 3. Using weighted average method, the above multiobjective programming becomes
min
_
(2 +4)x
2
1
+2( +3)x
1
x
2
+(9 +)x
2
2
+x
1
+(2 4)x
2
_
+(1 )
_
(7 )x
2
1
+2(5 )x
1
x
2
+(12 2)x
2
2
+(3 2)x
1
+(1 )x
2
_
,
subject to ( +1)x
1
+x
2
,
_
x
1
+x
2
_
+3,
(4 2)x
1
+(3 2)x
2
3 2,
(3 2)(x
1
+x
2
) 5 ,
x
1
0, x
2
0, 0 1.
(6.12)
Solution of this problem is tabulated for each value of by considering = 0.3 in
Table 6.1.
Here the solution is derived using the concept of fuzzy equivalence class and []-fuzzy
matrix. The fuzzy quadratic problem is solved for 11 aspiration levels and summarized
in the table. Each problem is solved by Lingo package. The advantage of this method,
compared to any defuzzication method, is that the solution corresponds to a particular
aspiration level.
18 Equivalence class in fuzzy numbers
Table 6.1
x
1
x
2
f (x)
0 0.5732861 0.2356186 4.565938
0.1 0.5700244 0.2263955 4.369359
0.2 0.5659454 0.2163832 4.163395
0.3 0.5608270 0.2054951 3.946557
0.4 0.5543727 0.1936398 3.717039
0.5 0.5461818 0.1807273 3.472718
0.6 0.5357056 0.1666801 3.211037
0.7 0.5221812 0.1514556 2.928989
0.8 0.5045319 0.1350882 2.623118
0.9 0.4812167 0.1177694 2.289659
1.0 0.4500000 0.1000000 1.925000
7. Conclusion
In this paper, the objective of introducing fuzzy equivalence class is to nd the member-
ship value of arithmetic operations of a large number of same type of fuzzy numbers and
to apply this in fuzzy decision-making problems. The introduction of []-fuzzy matrix is
to formulate fuzzy quadratic programming in which the matrix involved in the quadratic
form of the objective function has to be symmetric positive semidenite. Also this paper
deals with a solution method for fuzzy decision-making problem. The basic concept of
this method is to decompose the original fuzzy decision-making problem into two prob-
lems for a particular aspiration level. This methodology is applicable whenever all the
fuzzy numbers involved in the decision-making problem are of same type belonging to
one equivalence class. If the fuzzy numbers are of dierent type, that is, belonging to dif-
ferent equivalent classes, then such type of decomposition of the fuzzy decision-making
problem for an aspiration level will be dicult since in that case, the arithmetic operation
of fuzzy numbers is not closed. This diculty can be solved if a one-to-one correspon-
dence between two equivalence classes can be established. This is the future research scope
of this paper.
Acknowledgments
The authors would like to thank the referees for their valuable suggestions which im-
proved the presentation of the paper. The rst author is thankful to ISIRD, IIT Kharag-
pur for nancial support to prepare this paper. Research work of the second author is
supported by AICTE, India, and G. H. Patel College of Engineering and Technology, V. V.
Nagar-388120, Gujarat, India, and IIT Kharagpur.
References
[1] R. E. Bellman and L. A. Zadeh, Decision-making in a fuzzy environment, Management Science
17 (1970-1971), B141B164.
[2] D. Dubois and H. Prade, Operations on fuzzy numbers, International Journal of Systems Science
9 (1978), no. 6, 613626.
Geetanjali Panda et al. 19
[3] T.-S. Liou and M. J. J. Wang, Ranking fuzzy numbers with integral value, Fuzzy Sets and Systems
50 (1992), no. 3, 247255.
[4] M. Mizumoto and K. Tanaka, Algebraic properties of fuzzy numbers, Proceedings of the IEEE
International Conference on Cybernetics and Society, Washington, DC, 1976.
[5] R. E. Moore, Interval Analysis, Prentice-Hall, New Jersey, 1966.
[6] B. Werners, An interactive fuzzy programming system, Fuzzy Sets and Systems 23 (1987), no. 1,
131147.
[7] L. A. Zadeh, The concept of a linguistic variable and its application to approximate reasoning. I,
Information Science 8 (1975), no. 3, 199249.
[8] , The concept of a linguistic variable and its application to approximate reasoning. II, In-
formation Science 8 (1975), no. 4, 301357.
[9] H.-J. Zimmermann, Fuzzy mathematical programming, Computers & Operations Research and
Their Application to Problems of World Concern 10 (1983), no. 4, 291298.
Geetanjali Panda: Department of Mathematics, Indian Institute of Technology, Kharagpur-721302,
West Bengal, India
E-mail address: geetanjali@maths.iitkgp.ernet.in
Motilal Panigrahi: Department of Mathematics, Indian Institute of Technology, Kharagpur-721302,
West Bengal, India
E-mail address: motilal.panigrahi@gmail.com
Sudarsan Nanda: Department of Mathematics, Indian Institute of Technology, Kharagpur-721302,
West Bengal, India
E-mail address: snanda@maths.iitkgp.ernet.in
Mathematical Problems in Engineering
Special Issue on
Time-Dependent Billiards
Call for Papers
This subject has been extensively studied in the past years
for one-, two-, and three-dimensional space. Additionally,
such dynamical systems can exhibit a very important and still
unexplained phenomenon, called as the Fermi acceleration
phenomenon. Basically, the phenomenon of Fermi accelera-
tion (FA) is a process in which a classical particle can acquire
unbounded energy from collisions with a heavy moving wall.
This phenomenon was originally proposed by Enrico Fermi
in 1949 as a possible explanation of the origin of the large
energies of the cosmic particles. His original model was
then modied and considered under dierent approaches
and using many versions. Moreover, applications of FA
have been of a large broad interest in many dierent elds
of science including plasma physics, astrophysics, atomic
physics, optics, and time-dependent billiard problems and
they are useful for controlling chaos in Engineering and
dynamical systems exhibiting chaos (both conservative and
dissipative chaos).
We intend to publish in this special issue papers reporting
research on time-dependent billiards. The topic includes
both conservative and dissipative dynamics. Papers dis-
cussing dynamical properties, statistical and mathematical
results, stability investigation of the phase space structure,
the phenomenon of Fermi acceleration, conditions for
having suppression of Fermi acceleration, and computational
and numerical methods for exploring these structures and
applications are welcome.
To be acceptable for publication in the special issue of
Mathematical Problems in Engineering, papers must make
signicant, original, and correct contributions to one or
more of the topics above mentioned. Mathematical papers
regarding the topics above are also welcome.
Authors should follow the Mathematical Problems in
Engineering manuscript format described at http://www
.hindawi.com/journals/mpe/. Prospective authors should
submit an electronic copy of their complete manuscript
through the journal Manuscript Tracking System at http://
mts.hindawi.com/ according to the following timetable:
Manuscript Due December 1, 2008
First Round of Reviews March 1, 2009
Publication Date June 1, 2009
Guest Editors
Edson Denis Leonel, Department of Statistics, Applied
Mathematics and Computing, Institute of Geosciences and
Exact Sciences, State University of So Paulo at Rio Claro,
Avenida 24A, 1515 Bela Vista, 13506-700 Rio Claro, SP,
Brazil; edleonel@rc.unesp.br
Alexander Loskutov, Physics Faculty, Moscow State
University, Vorobevy Gory, Moscow 119992, Russia;
loskutov@chaos.phys.msu.ru
Hindawi Publishing Corporation
http://www.hindawi.com